{"created":"2023-05-15T12:22:23.520122+00:00","id":13582,"links":{},"metadata":{"_buckets":{"deposit":"3fdc9d8f-a1a3-4e12-a97c-1fae63bea607"},"_deposit":{"created_by":10,"id":"13582","owners":[10],"pid":{"revision_id":0,"type":"depid","value":"13582"},"status":"published"},"_oai":{"id":"oai:kansai-u.repo.nii.ac.jp:00013582","sets":["528:1588:1589:1591","63:69"]},"author_link":["34923","34918","34920","34919","34921","34924","34917","34922"],"item_12_alternative_title_20":{"attribute_name":"その他のタイトル","attribute_value_mlt":[{"subitem_alternative_title":"Learning of Object Concept and Application to Object Recognition Using Clustering and Logistic Regression"}]},"item_12_biblio_info_7":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicIssueDates":{"bibliographicIssueDate":"2018-08-15","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"8","bibliographicPageEnd":"1510","bibliographicPageStart":"1499","bibliographicVolumeNumber":"59","bibliographic_titles":[{"bibliographic_title":"情報処理学会論文誌 "}]}]},"item_12_description_4":{"attribute_name":"概要","attribute_value_mlt":[{"subitem_description":"精度が高い物体識別方法としてCNN (Convolutional Neural Networks) が一般的に利用されている．しかし，この手法は各カテゴリの学習データとして数万枚の画像が必要であり，膨大な学習時間が必要である．またこの手法はどのような特徴を利用して識別を行っているかが分からない．実際に，高い識別率を持つCNNでも，人間の直感から外れた認識をすることが報告されている．一方で，人間はあるカテゴリの画像を数枚見るだけで，物体の概念を得ることが可能である．さらに人間は言葉で概念を表現することも可能である．本研究では，クラスタリングとロジスティック回帰を利用した物体概念の学習方法を提案する．提案手法は複数の低次元な特徴（具体的には，色，輪郭，大きさ）を利用することで，物体概念の生成を行い，高い識別率と可読性を持つ識別器を作成する．提案手法の有効性を，RGB-Dオブジェクトデータセットを用いた従来手法との比較により実証した． ","subitem_description_type":"Other"},{"subitem_description":"In general, CNN (Convolutional Neural Networks) is used as the method with high recognition accuracy. In CNN, however, several tens of thousands images are required as learning data for each category. Also, huge learning time is required. Another drawback is that we cannot understand what CNN focuses the features when recognizing, in fact, it has been reported that well-trained CNN has recognition results that are out of human intuition. In contrast, after a human just look at several objects in a category he can get something like its general object concept. Furthermore, a human can represent the concept by words. In this article, a new concept learning method based on clustering and logistic regression is proposed. The proposed method learns object concepts from multiple low-dimensional features, e.g., color, contour, and size based on which it generates classifier with high accuracy and readability. Effectiveness of the proposed method was demonstrated by comparison with the conventional method using RGB-D Object Dataset. ","subitem_description_type":"Other"}]},"item_12_description_5":{"attribute_name":"内容記述","attribute_value_mlt":[{"subitem_description":"本研究の一部は、平成28年度関西大学教育研究高度促進費において、研究課題「ロボット競技会をモチベーションとしたソフトウェアに力点をおいたメカトロニクス教育」として研究費を受けた。","subitem_description_type":"Other"}]},"item_12_full_name_3":{"attribute_name":"著者別名","attribute_value_mlt":[{"nameIdentifiers":[{"nameIdentifier":"34921","nameIdentifierScheme":"WEKO"}],"names":[{"name":"Akimoto , Shohei"}]},{"nameIdentifiers":[{"nameIdentifier":"34922","nameIdentifierScheme":"WEKO"}],"names":[{"name":"Takahashi , Tomokazu"}]},{"nameIdentifiers":[{"nameIdentifier":"34923","nameIdentifierScheme":"WEKO"}],"names":[{"name":"Suzuki , Masato"}]},{"nameIdentifiers":[{"nameIdentifier":"34924","nameIdentifierScheme":"WEKO"}],"names":[{"name":"Aoyagi , Seiji"}]}]},"item_12_publisher_34":{"attribute_name":"出版者","attribute_value_mlt":[{"subitem_publisher":"情報処理学会"}]},"item_12_rights_13":{"attribute_name":"権利","attribute_value_mlt":[{"subitem_rights":"©2018 Information Processing Society of Japan : 利用は著作権の範囲内に限る。"}]},"item_12_source_id_10":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AN00116647","subitem_source_identifier_type":"NCID"}]},"item_12_source_id_8":{"attribute_name":"ISSN","attribute_value_mlt":[{"subitem_source_identifier":"18827764","subitem_source_identifier_type":"ISSN"}]},"item_12_version_type_17":{"attribute_name":"著者版フラグ","attribute_value_mlt":[{"subitem_version_resource":"http://purl.org/coar/version/c_970fb48d4fbd8a85","subitem_version_type":"VoR"}]},"item_creator":{"attribute_name":"著者","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"秋本, 翔平 "}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"高橋, 智一 "}],"nameIdentifiers":[{},{}]},{"creatorAffiliations":[{"affiliationNameIdentifiers":[{"affiliationNameIdentifier":"","affiliationNameIdentifierScheme":"ISNI","affiliationNameIdentifierURI":"http://www.isni.org/isni/"}],"affiliationNames":[{"affiliationName":"","affiliationNameLang":"ja"}]}],"creatorNames":[{"creatorName":"鈴木, 昌人 ","creatorNameLang":"ja"},{"creatorName":"Suzuki, Masato","creatorNameLang":"en"}],"familyNames":[{"familyName":"鈴木","familyNameLang":"ja"},{"familyName":"Suzuki","familyNameLang":"en"}],"givenNames":[{"givenName":"昌人 ","givenNameLang":"ja"},{"givenName":"Masato","givenNameLang":"en"}],"nameIdentifiers":[{},{}]},{"creatorNames":[{"creatorName":"青柳, 誠司 "}],"nameIdentifiers":[{},{}]}]},"item_files":{"attribute_name":"ファイル情報","attribute_type":"file","attribute_value_mlt":[{"accessrole":"open_date","date":[{"dateType":"Available","dateValue":"2019-10-16"}],"displaytype":"detail","filename":"KU-0020-20180815-00.pdf","filesize":[{"value":"2.1 MB"}],"format":"application/pdf","licensetype":"license_note","mimetype":"application/pdf","url":{"label":"KU-0020-20180815-00.pdf","url":"https://kansai-u.repo.nii.ac.jp/record/13582/files/KU-0020-20180815-00.pdf"},"version_id":"a5b35b61-6783-433a-a6d5-5d3b8c77d145"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"物体認識","subitem_subject_scheme":"Other"},{"subitem_subject":"物体概念","subitem_subject_scheme":"Other"},{"subitem_subject":"RGB-Dセンサ","subitem_subject_scheme":"Other"},{"subitem_subject":"クラスタリング","subitem_subject_scheme":"Other"},{"subitem_subject":"ロジスティック回帰","subitem_subject_scheme":"Other"},{"subitem_subject":"object recognition","subitem_subject_scheme":"Other"},{"subitem_subject":"object concept","subitem_subject_scheme":"Other"},{"subitem_subject":"RGB-D sensor","subitem_subject_scheme":"Other"},{"subitem_subject":"clustering","subitem_subject_scheme":"Other"},{"subitem_subject":"logistic regression","subitem_subject_scheme":"Other"},{"subitem_subject":"関西大学","subitem_subject_scheme":"Other"},{"subitem_subject":"Kansai University","subitem_subject_scheme":"Other"},{"subitem_subject":"関西大学教育研究高度化促進費","subitem_subject_scheme":"Other"}]},"item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"jpn"}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourcetype":"research report","resourceuri":"http://purl.org/coar/resource_type/c_18ws"}]},"item_title":"クラスタリングとロジスティック回帰を利用した物体概念の学習と認識への応用","item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"クラスタリングとロジスティック回帰を利用した物体概念の学習と認識への応用"}]},"item_type_id":"12","owner":"10","path":["69","1591"],"pubdate":{"attribute_name":"公開日","attribute_value":"2019-10-16"},"publish_date":"2019-10-16","publish_status":"0","recid":"13582","relation_version_is_last":true,"title":["クラスタリングとロジスティック回帰を利用した物体概念の学習と認識への応用"],"weko_creator_id":"10","weko_shared_id":-1},"updated":"2023-11-29T05:48:13.221516+00:00"}